Web Semantics: Cutting Edge and Future Directions in Healthcare

By Vishal Jain and Valentina Emilia Balas

Web Semantics strengthen the description of web resources to exploit them better and make them more meaningful for both humans and machines, thereby contributing to the development of a knowledgeintensive data web. The world is experiencing the movement of concept from data to knowledge and the movement of web from document model to data model. The underlying idea is making the data machine understandable and processable. In the light of these trends, conciliation of Semantic and the Web is of paramount importance for further progress in the area. Web Semantics: Cutting Edge and Future Directions in Healthcare describes the three major components of the study of Semantic Web, namely Representation, Reasoning, and Security with a special focus on the healthcare domain. This book summarizes the trends and current research advances in web semantics, emphasizing the existing tools and techniques, methodologies, and research solutions. It provides easily comprehensible information on Web Semantics including semantics for data and semantics for services.

• Presents a comprehensive examination of the emerging research in areas of the semantic web, including ontological engineering, semantic annotation, reasoning and intelligent processing, semantic search paradigms, semantic web mining, and semantic sentiment analysis
• Helps readers understand key concepts in semantic web applications for biomedical engineering and healthcare, including mapping disparate knowledge bases, security issues, multilingual semantic web, and integrating databases with knowledge bases
• Includes coverage of key application areas of the semantic web, including clinical decision-making, biodiversity science, interactive healthcare, intelligent agent systems, decision support systems, and clinical natural language processing

• Language: English
• Publisher: Academic Press
• Release date: Mar 27, 2021
• ISBN: 9780128224854

Book preview

Germany

Preface

Sarika Jain (India), Vishal Jain (India) and Valentina Emilia Balas (Romania)

Over the last decade, we have witnessed an increasing use of Web Semantics as a vital and ever-growing field. It incorporates various subject areas contributing to the development of a knowledge-intensive data web. In parallel to the movement of concept from data to knowledge, we are now also experiencing the movement of web from document model to data model where the main focus is on data compared to the process. The underlying idea is making the data machine understandable and processable. In light of these trends, conciliation of Semantic and the Web is of paramount importance for further progress in the area. The 17 chapters in this volume, authored by key scientists in the field are preceded by an introduction written by one of the volume editors, making a total of 18 chapters. Chapter 1, Introduction, by Sarika Jain provides an overview of technological trends and perspectives in Web Semantics, defines Semantic Intelligence, and discusses the technologies encompassing the same in view of their application within enterprises as well as in web. In all, 76 chapter proposals were submitted for this volume making a 22% acceptance rate. The chapters have been divided into three sections as Representation, Reasoning, and Security.

• Representation: The semantics have to be encoded with data by virtue of technologies that formally represent metadata. When semantics are embedded in data, it offers significant advantages for reasoning and interoperability.

• Reasoning: When Semantic Web will finally happen, machine will be able to talk to machines materializing the so-called intelligent agents. The services offered will be useful for web as well as for the management of knowledge within an organization.

• Security: In this new setting, traditional security measures will not be suitable anymore; and the focus will move to trust and provenance. The semantic security issues are required to be addressed by the security professionals and the semantic technologists.

This book will help the instructors and students taking courses of Semantic Web getting abreast of cutting edge and future directions of semantic web, hence providing a synergy between healthcare processes and semantic web technologies. Many books are available in this field with two major problems. Either they are very advanced and lack providing a sufficiently detailed explanation of the approaches, or they are based on a specific theme with limited scope, hence not providing details on crosscutting areas applied in the web semantic. This book covers the research and practical issues and challenges, and Semantic Web applications in specific contexts (in this case, healthcare). This book has varied audience and spans industrial professionals, researchers, and academicians working in the field of Web Semantics. Researchers and academicians will find a comprehensive study of the state of the art and an outlook into research challenges and future perspectives. The industry professionals and software developers will find available tools and technologies to use, algorithms, pseudocodes, and implementation solutions. The administrators will find a comprehensive spectrum of the latest viewpoint in different areas of Web Semantics. Finally, lecturers and students require all of the above, so they will gain an interesting insight into the field. They can benefit in preparing their problem statements and finding ways to tackle them.

The book is structured into three sections that group chapters into three otherwise related disections:

Representation

The first section on Representation comprises six chapters that specifically focus on the problem of choosing a data model for representing and storage of data for the Web. Chapter 2, Convology: an ontology for conversational agents in digital health by Dragoni et al. propose an ontology, namely, Convology, aiming to describe conversational scenarios with the scope of providing a tool that, once deployed into a real-world application, allows to ease the management and understanding of the entire dialog workflow between users, physicians, and systems. The authors have integrated Convology into a living lab concerning the adoption of conversational agents for supporting the self-management of patients affected by asthma. Dubey et al. in Chapter 3, Conversion between semantic data models: the story so far, and the road ahead, provide the trends in converting between various semantic data models and reviews the state of the art of the same. In Chapter 4, Semantic interoperability: the future of healthcare Burse et al. have beautifully elaborated the syntactic and semantic interoperability issues in healthcare. They have reviewed the various healthcare standards in an attempt to solve the interoperability problem at a syntactic level and then moves on to examine medical ontologies developed to solve the problem at a semantic level. The chapter explains the features of semantic web technology that can be leveraged at each level. A literature survey is carried out to gage the current contribution of semantic web technologies in this area along with an analysis of how semantic web technologies can be improved to better suit the health-informatics domain and solve the healthcare interoperability challenge. Haklae Kim in his Chapter 5, A knowledge graph of medical institutions in Korea, has proposed a knowledge model for representing medical institutions and their characteristics based on related laws. The author also constructs a knowledge graph that includes all medical institutions in Korea with an aim to enable users to identify appropriate hospitals or other institutions according to their requirements. Chapter 6, Resource description framework based semantic knowledge graph for clinical decision support systems, by Lourdusamy and Mattam advocates the use of Semantic Knowledge Graphs as the representation structure for Clinical Decision Support Systems. Patnaikuni and Gengaje in Chapter 7, Probabilistic, syntactic, and semantic reasoning using MEBN, OWL, and PCFG in healthcare, exploit the key concepts and terminologies used for representing and reasoning uncertainties structurally and semantically with a case study of COVID-19 Corona Virus. The key technologies are Bayesian networks, Multi-Entity Bayesian Networks, Probabilistic Ontology Web Language, and probabilistic context-free grammars.

Reasoning

At the scale of www, logic-based reasoning is not appropriate and poses numerous challenges. As already stated in different chapters of Section 1, RDF provides a machine-processable syntax to the data on the web. Reasoning on Semantic Web involves deriving facts and relationships that are not explicit in the knowledge base. This section groups 10 contributions based on reasoning within the knowledge bases. There is an absence of a reference model for describing the health data and their sources and linking these data with their contexts. Chapter 8, The connected electronic health record: a semantic-enabled, flexible, and unified electronic health record, by Sassi and Chbeir addresses this problem and introduces a semantic-enabled, flexible, and unified electronic health record (EHR) for patient monitoring and diagnosis with Medical Devices. The approach exploits semantic web technologies and the HL7 FHIR standard to provide semantic connected EHR that will facilitate data interoperability, integration, information search and retrieval, and automatic inference and adaptation in real-time. Jain et al. in Chapter 9, Ontology-supported rule-based reasoning for emergency management, have proposed an ontology-supported rule-based reasoning approach to automate the process of decision support and recommending actions faster than a human being and at any time. Chapter 10, Healthcare-Cube Integrator for Healthcare Databases by Trivedi et al. proposes the Healthcare-cube integrator as a knowledge base that is storing health records collected from various healthcare databases. They also propose a processing tool to extract data from assorted databases. Chapter 11, Smart mental healthcare systems, by Dalal and Jain provides an architecture for a smart mental healthcare system along with the challenges and benefits incurred. Chapter 12, A meaning-aware information search and retrieval framework for healthcare, by Anoop et al. discusses a framework for building a meaning-aware information extraction from unstructured EHRs. The proposed framework uses medical ontologies, a medical catalog-based terminology extractor and a semantic reasoner to build the medical knowledge base that is used for enabling a semantic information search and retrieval experience in the healthcare domain. In Chapter 13, Ontology-based intelligent decision support systems: a systematic approach, Saha et al. emphasize several machine learning algorithms and semantic technologies to design and implement intelligent decision support system for effective healthcare support satisfying quality of service and quality of experience requirements. Jacyntho and Morais in Chapter 14, Ontology-based decision-making, have described the architecture and strengths of knowledge-based decision support systems. They have defined a method for the creation of ontology-based knowledge bases and a corresponding fictitious health care case study but with real-world challenges. As the data are exploding over the web, Daoui et al. in Chapter 15, A new method for profile identification using ontology-based semantic similarity, aim to treat and cover a new system in the domain of tourism in order to offer users of the system a set of interesting places and tourist sites according to their preferences. The authors focus on the design of a new profile identification method by defining a semantic correspondence between keywords and the concepts of an ontology using an external resource WordNet. Compared to the objective type assessment, the descriptive assessment has been found to be more uniform and at a higher level of Bloom’s taxonomy. In Chapter 16, Semantic similarity-based descriptive answer evaluation, Shaukat et al. have put in efforts to deal with the problem of automated computer assessment in the descriptive examination. Lastly in this section, Chapter 17, Classification of genetic mutations using ontologies from clinical documents and deep learning, by Bedi et al. have presented a framework for classifying cancerous genetic mutation reported in EHRs. They have utilized clinical NLP, Ontologies and Deep Learning for the same over Catalog of Somatic Mutations in Cancer Mutation data and Kaggle’s cancer-diagnosis dataset.

Security

Though posed as the future of web, is semantic web secure? In the semantic web setting, traditional security measures are no more suitable. This section closes the book by providing Chapter 18, Security issues for the semantic web, by Pranav et al. providing the security issues in the semantic web. This chapter also suggested ways of potentially aligning the protocols so as to make them more robust to be used for semantic web services.

As the above summary shows, this book summarizes the trends and current research advances in web semantics, emphasizing the existing tools and techniques, methodologies, and research solutions.

Chapter 1

Semantic intelligence: An overview

Sarika Jain,    Department of Computer Applications, National Institute of Technology Kurukshetra, Haryana, India

Abstract

The web is ubiquitously influencing all aspects of commerce and society, as well as information processing and dissemination. With the emergence of the semantic web, it is moving toward application and knowledge deployment. Richer, meaningful representations offer more insight, powerful reasoning capabilities, and enhanced security. Web Semantics strengthen the description of web resources for exploiting them better and making them more meaningful for both human and machine.

Keywords

Web Semantics; Semantic Intelligence; Semantic technologies; ontology; publishing data; consuming data

1.1 Overview

Due to many technological trends like IoT, Cloud Computing, Smart Devices, huge data is generated daily and at unprecedented rates. Traditional data techniques and platforms do not prove to be efficient because of issues concerning responsiveness, flexibility, performance, scalability, accuracy, and more. To manage these huge datasets and to store the archives for longer periods, we need granular access to massively evolving datasets. Addressing this gap has been an important and well-recognized interdisciplinary area of Computer Science.

A machine will behave intelligently if the underlying representation scheme exhibits knowledge that can be achieved by representing semantics. Web Semantics strengthen the description of web resources for exploiting them better and making them more meaningful for both human and machine. As semantic web is highly interdisciplinary, it is emerging as a mature field of research that facilitates information integration from variegated sources. Semantic web converts data to meaningful information and is therefore a web of meaningful, linked, and integrated data by virtue of metadata. Current web is composed primarily of unstructured data, such as HTML pages and search in current web is based on keyword search. These searches are not able to make out the type of information on the HTML page, that is, it is not possible to extract different pieces of data from different web pages about a concept and then give integrated information about the concept. The semantic web provides such a facility with lesser human involvement.

As the web connects documents, in the same manner, semantic web connects pieces of information. In addition to publishing data on the World Wide Web, the semantic web is being utilized in enterprises for myriad of use cases. The Artificial Intelligence technologies, the Machine Intelligence technologies, and the semantic web technologies together make up the Semantic Intelligence technologies (SITs). SITs have been found as the most important ingredient in building artificially intelligent knowledge-based systems as they aid machines in integrating and processing resources contextually and intelligently.

This book describes the three major compartments of the study of Web Semantics, namely representation, reasoning, and security. It also covers the issues related to the successful deployment of semantic web. This chapter addresses the key knowledge and information needs of the audience of this book. It provides easily comprehensible information on Web Semantics including semantics for data and semantics for services. Further, an effort has been made to cover the innovative application areas semantic web goes hand in hand with a focus on Health Care.

1.2 Semantic Intelligence

Semantic Intelligence refers to filling the semantic gap between the understanding of humans and machines by making a machine look at everything in terms of object-oriented concepts as a human look at it. Semantic Intelligence helps us make sense of the most vital resource, that is, data; by virtue of making it interpretable and meaningful. The focus is on information as compared to the process. To whatever application, the data will be put to; it is to be represented in a manner that is machine-understandable and hence human-usable. All the important relationships (including who, what, when, where, how, and why) in the required data from any heterogeneous data source are required to be made explicit.

The primary technology standards of the SITs are RDF (Resource Description Framework) and SPARQL (SPARQL Protocol and RDF Query Language). RDF is the data model/format/serialization used to store data. SPARQL is the query language designed to query, retrieve, and process data stored as RDF across various systems and databases. Both of these technologies are open-ended making them a natural fit for iterative, flexible, and adaptable software development in a dynamic environment; hence suitable for a myriad of open-ended problems majorly including unstructured information. It is even beneficial to wrap up the existing relational data stores with the SPARQL end points to integrate them with any intelligent application. This all is possible because semantic web operates on the principle of Open World Assumption; wherein all the facts are not anticipated in the beginning; and in the absence of some fact, it cannot be assumed false.

Semantics is no more than discovering relationships between things. These relationships when discovered and represented explicitly help manage the data more efficiently by making sense of it. In addition to storing and retrieving information, semantic intelligence provides a flexible model by acting as an enabler for machines to infer new facts and derive new information from existing facts and data. In all such systems with a large amount of unstructured and unpredictable data, SITs prove to be less cost-intensive and maintainable. By virtue of being able to interpret all the data, machines are able to perform sophisticated tasks for the mankind. In today’s world SITs are serving a very broad range of applications, across multiple domains, within enterprises, and on the web. A full-fledged industry in its own sense has emerged in the last 20 years when these technologies were merely drafts. In addition to publishing and consuming data on the web, SITs are being used in enterprises for various purposes.

1.2.1 Publishing and consuming data on the web

Publishing data on the web involves deciding upon the format and the schema to use. Best practices exist to publish, disseminate, use, and perform reasoning on high-quality data over the web. RDF data can be published in different ways including the linked data (DBPedia), SPARQL endpoint, metadata in HTML (SlideShare, LinkedIn, YouTube, Facebook), feeds, GRDDL, and more. Semantic interlinked data is being published on the web in all the domains including e-commerce, social data, and scientific data. People are consuming this data through search engines and specific applications. Publishing semantic web data about the web pages, an organization ensures that the search results now also include related information like reviews, ratings, and pricing for the products. This added information in search results does not increase ranking of a web page but significantly increases the number of clicks this web page can get. Here are some popular domains where data is published and consumed on the semantic web.

• E-commerce: The Schema.org and the GoodRelations vocabulary are global schema for commerce data on the web. They are industry-neutral, syntax-neutral, and valid across different stages of value chain.

• Health care and life sciences: HealthCare is a novel application domain of semantic web that is of prime importance to human civilization as a whole. It has been predicted as the next big thing in personal health monitoring by the government. Big pharma companies and various scientific projects have published a significant amount of life sciences and health care data on the web.

• Media and publishing: The BBC, The FT, SpringerNature, and many other media and publishing sector companies are benefitting their customers by providing an ecosystem of connected content to provide more meaningful navigation paths across the web.

• Social data: A social network is a two-way social structure made up of individuals (persons, products, or anything) and their relationships. The Facebook’s social graph represents connections between people. Social networking data using friend-of-a-friend as vocabulary make up a significant portion of all data on the web.

• Linked Open Data: A powerful data integration technology is the practical side of semantic web. DBPedia is a very large-linked dataset making the content of Wikipedia available to the public as RDF. It incorporates links to various other datasets as Geonames; thus allowing applications to exploit the extra and more precise knowledge from other datasets. In this manner, applications can provide a high user experience by integrating data from multiple linked datasets.

• Government data: For the overall development of the society, the governments around the world have taken initiatives for publishing nonpersonal data on the web making the government services transparent to the public.

1.2.2 Semantic Intelligence technologies applied within enterprises

Enterprise information systems comprise complex, distributed, heterogeneous, and voluminous data sources. Enterprises are leveraging SITs to achieve interoperability and implement solutions and applications. All documents are required to be semantically tagged with the associated metadata.

• Information classification: The knowledge bases as are used by the giants Facebook, Google, and Amazon today are said to shape up and classify data and information in the same manner as the human brain does. Along with data, a knowledge base also contains expert knowledge in the form of rules transforming this data and information into knowledge. Various organizations represent their information by combining the expressivity of ontologies with the inference support.

• Content management and situation awareness: The organizations reuse the available taxonomic structures to leverage their expressiveness to enable more scalable approaches to achieve interoperability of content.

• Efficient data integration and knowledge discovery: The data is scaling up in size giving rise to heterogeneous datasets as data silos. The semantic data integration allows the data silos to be represented, stored, and accessed using the same data model; hence all speaking the same universal language, that is, SITs. The value of data explodes when it is linked with other data providing more flexibility compared to the traditional data integration approaches.

1.3 About the book

This book contains the latest cutting-edge advances and future directions in the field of Web Semantics, addressing both original algorithm development and new applications of semantic web. It presents a comprehensive up-to-date research employing semantic web and its health care applications, providing a critical analysis of the relative merit, and potential pitfalls of the technique as well as its future outlook.

This book focuses on a core area of growing interest, which is not specifically or comprehensively covered by other books. This book describes the three major compartments of the study of Web Semantics, namely Representation, Reasoning, and security. It covers the issues related to the successful deployment of semantic web. Further, an effort has been made to cover the innovative application areas semantic web goes hand in hand with focus on HealthCare by providing a separate section in every chapter for the case study of health care, if not explicitly mentioned. The book will help the instructors and students taking courses of semantic web getting abreast of cutting edge and future directions of semantic web, hence providing a synergy between health care processes and semantic web technologies.

Section I

Representation

Outline

Chapter 2 Convology: an ontology for conversational agents in digital health

Chapter 3 Conversion between semantic data models: the story so far, and the road ahead

Chapter 4 Semantic interoperability: the future of healthcare

Chapter 5 A knowledge graph of medical institutions in Korea

Chapter 6 Resource description framework based semantic knowledge graph for clinical decision support systems

Chapter 7 Probabilistic, syntactic, and semantic reasoning using MEBN, OWL, and PCFG in healthcare

Chapter 2

Convology: an ontology for conversational agents in digital health

Mauro Dragoni¹, Giuseppe Rizzo² and Matteo A. Senese²,    ¹Fondazione Bruno Kessler, Trento, Italy,    ²LINKS Foundation, Torino, Italy

Abstract

Conversational agents are a modality for making the human–computer interaction paradigm more friendly from the user perspective. Conversational agents rely on natural language understanding capabilities for classifying the intents that users want to communicate through open natural language text. Recently, conversational agents are equipped with background knowledge for improving the overall effectiveness, efficiency, and reliability of systems concerning the acquisition of information from the dialog management perspective. However, while the literature discussed some introductory strategies, there are no evidence that such knowledge-equipped conversational agents have been used in practice. Within the digital health domain, the use of conversational agents ranges from assisting patients during the self-management of chronic diseases to supporting physicians during daily activities. In this chapter, we propose an ontology, namely Convology, aiming to describe conversational scenarios with the scope of providing a tool that, once deployed into a real-world application, allows to ease the management and understanding of the entire dialog workflow between users, physicians, and systems. We integrated Convology into a living laboratory concerning the adoption of conversational agents for supporting the self-management of patients affected by asthma. Observer results demonstrated the feasibility of investigating this research direction.

Keywords

Science; publication; complicated

2.1 Introduction

The conversation paradigm has been implemented for the realization of conversational agents overwhelmingly in the last years. Natural and seamless interactions with automated systems introduce a shift from using well-designed and sometimes complicated interfaces made of buttons and paged procedures to textual or vocal dialogs. Asking questions naturally has many advantages with respect to traditional app interactions. The main one is that the user does not need to know how the specific application works, everyone knows how to communicate, and in this case, the system is coming toward the user to make the interaction more natural. This paradigm has been integrated into mobile applications for supporting users from different perspectives and into more well-known systems built by big tech players like Google Assistant and Amazon Alexa. These kinds of systems dramatically reduce the users’ effort for asking and communicating information to systems that, by applying natural language understanding (NLU) algorithms, are able to decode which are the actual users’ intentions and to reply properly. However, by performing a deeper analysis of these systems, we can observe a strong limitation of their usage into complex scenarios. The interactions among users and bots are often limited to a single-turn communication where one of the actor sends an information request (e.g., a question like How is the weather today in London? or a command like Play the We Are The Champions song) and the other actor provides an answer containing the required information or performs the requested action (e.g., Today the weather in London is cloudy. or the execution of the requested